13" World Conference on Earthquake Engineering
a3 eee Vancouver, B.C., Canada
P August 1-6, 2004
se Paper No. 2306
apn 28
IEEE 693 SEISMIC QUALIFICATION OF COMPOSITES FOR
SUBSTATION HIGH-VOLTAGE EQUIPMENT
Anshel J. SCHIFF! and Leon KEMPNER, J
SUMMARY
‘This paper discusses research on the application of IEEE Standard 693 seismic qualification
requirements for hollow-core composite insulators used on high-voliage instrument transformer
substation equipment. The IEEE 693 qualification procedure of time history shake-table and static-
pall tests as well asthe acceptance criteria is evaluated. Failure modes for composite insulators are
discussed. Experimental observations show that the acceptance criteria are invalid tor qualifying,
hollow-core composites. The need for a static-pull test following vibration qualification tests is
presented. The need for post-vibration-test measurement of damping is discussed. Changes to the
IEEE standard and potcntial research are recommended
INTRODUCTION
IEEE Standard 693-1997, "IEEE Recommended Practice for Seismic Design of Substations" [1]. is
12 major improvement in the way the power industry seismically qualifies substation high-voltage
equipment. The standard has established discrete qualification levels with associated controlling
response spectra, and well-defined qualification procedures and acecptance criteria. In addition to
general procedures for analysis and testing, special procedures and criteria are developed for
vidual classes of equipment that reflect their unique characteristics and carthquake
performance.
“The standard provides twa scismic qualification performance levels that are defined by excitation
response spectra. Acceptance criteria are established so that equipment shake table-tests
qualification need only be subjected to an excitation half of those defined by the performance
Response spectra. While not explicitly stated in the standard, it is the view of the authors that
testing to half of the performance level requires that the failure modes of the equipment are
understood and that associated critical variables, such as a strain, can be measured. The acceptance
* Consulting Engincer, Precision Measurement Instruments, 27750 Edgerton Rd, Los Altos Hills,
CA, USA, Email: schiff@ee.stanford.edu
BPA, P.O. Box 61409 (TNFC-TPP3), Vancouver, WA 98666-1409, Principal Structural
Engincer, Bonneville Power Administration, Vancouver, WA, USA. Email: Ikempnerjr@ bpa.govcriteria require that the measured values of the critical variables be half their ultimate value. ‘This
procedure also assumes that the critical variables that are monitored are approximately linear. The
current standard does not explicitly specify the long-ierm serviceability of the equipment after
surviving an input excitation at the qualification level, but it is generally expected that it not be
affected.
Special tests and acceptance critcria arc established for fiber-wound polymer-impregnated
insulators, These are referred to as composite insulators, Composite insulators of the hollow-core
designs are used for post insulsiors, instrument transformers, surge arresters, and bushings. The
hollow-core type composite design is the subject of this paper. The composite tube is fitted with
‘metal flanges at its ends and silicon-rubber sheds are applied to the outside of the composite tube.
For some application identified above, the tube assembly is supplied to power equipment
manufacturers where it is used to fabricate the final product, such as an instrument transformer.
DESIGN AND FABRICATION OF HOLLOW-CORE COMPOSITE INSULATORS
‘The structural designs used by some manufacturers are equivalent although details of fabrication
may vary. A composite manufacturer may provide a tube and flange fittings ta meet the
specifications of the power equipment manufacturer using the composite. Other power equipment
manufacturers may he developing or using composite components of their own design and these
may be different from those described below.
It is useful ta know the general design and fabrication method of hollow-care composites in
understanding qualification tests, acceptance criteria, and potential failure modes. For many
applications the composite tube is a constant diameter circular cylinder. For some bushings the
tubes are tapered circular cylinders. Some of the variables involved in fabricating the tube include
the diameter of the tube. its wall thickness, the angle used to wind the fiber. and the thickness and
number of fiber layers. There may also be variation in the properties of the fiber and polymer used
in fabrication. The tube is wound on a mandrel and its outside diameter is carefully controlled to fit
its end flanges. After the tube is fabricated, it is cut to length... The metal end assembly is
illustrated in Figure 1. This figure does not show the shed material that is added after the flange is
attached to the tube. It should be noted that this illustration shows an end fitting with guscets, but
some manufacturers have designs without gussets. Also, the number off connection balis can vary
and are typically more numerous than illustrated. Important end-flange parameters include the
inside diameter to assure a good fit to the outside diameter of the tube, wall thickness, and the
length that the tube engages the flange (penetration length). The end of the flange may be open so
‘components can be inserted into the tube when fabricating electrical equipment, such as an
instrument transformer.
Prior to assembling the tube to the flange, the flange is heated so that it expands. A bonding
‘compound is applied to the tube-or flange. and the tube is inserted into the heated flange. When the
flange cools, it shrinks providing 4 compressive fit to the tube, Thus, two mechanisms are at play
in the tube-flange connection, a compressive stress due to the shrink fit of the flange to the tube and
the honding of the tmbe-flange interface surface.Section A-A
Figure I Schematic diagram of section of bollow-core composite flange assembly
SEISMIC STRUCTURAL FAILURE MODES.
Four failure modes are discussed: bond degradation, bond failure, tube degradation, we layer
delamination. These failure modes are associated with the composite insulator. Other system
failure modes can be associated with the deformation or failure of the flange, the failure of the
‘connection bolts, or the failure of the member to-which the composite insulator is attached, such as
the lid of an instrument-ransformer bax. The failure of the shed seal is not considered here. These
failure modes have been observed over the historic development of hollow-core campasites, but
may not be observed using current designs under seismic qualification or earthquake loads.
‘The most severe load that can cause failure is a laicral load during the qualification testing process
‘or during an canthquake. This subjects the lower flange-tube connection to large bending moments.
It is interesting to note the difference in the scismic performance of porcelain and composite
insulators, When poreetain fails, the failure is catastrophic and obvious. As discussed below,
‘composite damage is generally incremental and in many cases non-caiastrophic damage may be
difficult to detect.
Bond Degradation
When a round tube is subjected to bending. it tends to deform into an oval shape. This deformation
and the reaction forces introduced by a Iateral load to the tube will subject the tube-flange interface
bond opposite the direction of the load to a peeling force and shear force. Typically, peeling forces
are most severe to a bonded connection, as high stress is concentrated along a line rather than over
an area subjected to the shear forces. A local bond separation near the upper edge of the flange will
have several effects. The shear forees on the bond that remains are increased and the stiffness of
the connection is reduced. When subjected to vibration it is possible for that part of the bond that
has separated to be mobilized. that is, there can be relative motion across the separated surface of
the bond. This can introduce additional damping torthe system. 1t should be noted that there is still
‘compressive stress om the connection duc to the shrink fit uscd to assemble the connection. A
partial bond separation, or hond degradation, will not jeopardize the integrity of the seal between
the tube and the flange if it does not extend over the full length of the bond.Severe tond Degradation
I the bond degradation progresses, two types of failure can occur. The bond separation can extend
to the base of the flange so that the seal between the tube and flange can be compromised, If the
entire bond fails, the tube can partially pull out of the flange: and break the silicon rubber that
‘covers ihe top of the flange as shown in Figure 2.
Figure 2 Bond failure and tube partially pulled from flange breaking shed rubber
‘Tube Degradation
When the polymer-impregnated fiber is subjected to bending or any load, some fibers will break.
‘even at very low stress levels. A small number of fiber breakage would not be considered damage.
‘As loads increase the number of fibers that break increases and at some point the tube would be
‘considered damaged. As the load increases, eventually a hinge will form and a large displacement
response of the tube at the hinge would be obvious. It is this condition that the current standard
criteria address. Before a hinge develops, the effects of fiber breakage will be a reduction in
stiffness of the composite member and in increase in damping. The silicon-rubber sheds that cover
the composite will obscure direct ebservation of the damaged composite material.
‘Tube Layer Delamination
In fabricating the tube, a laycr of fibers is put down with the fibers in a given direction followed by
another layer of fibers in a skewed direction. When subject to stress. a crack can form between
these layers and this is referred to as delamination, While this will result in some reduction in
strength, a more significant problem is that electrical corona can develop across the crack when the
units energized and eventually cause an electrical failure.
QUALIFICATION PROCEDURES FOR IEEE 693
‘The procedure for qualifying composite insulators requires that the insulator stack be anchored to.a
stiff support and be subjected to a cantilever pull test to 1/2 the Specific Mechanical Load (SML).
‘The SML js the manutacturer-specified load the insulator will withstand without visual damage, but
that is above the damage limit zone. The 1/2 SML. is below the damage threshold for the item being,
tested. During this test. the relative deflection at the top of the stack is measured and this deflection
is used as the acceptance criteria for the shake-table test. When the 1/2 SML is removed, the
residual deflection. that is, the deflection remaining when the load is removed, is measured. This
deflection must be less than 5% of the peak deflection observed at the U2 SML. During shake-
table tests, the standard allows the relative deflection to be measured directly with wire
potentiometers or indirectly hy double integrating the accelerations at the top and bottom of the
‘composite member.
“The test to determine the SML. which is done by the tube manufacturer independently of the
‘equipment qualification test, is mich different than that used to determine the ultimate strength of aporcelain member. The load that causes failure of composite membcr is a function of the duration
that the load is applied. The SML is the load that causes the composite insulator to fail when the
oad is applied far onc minute. A composite will fail if it is subjected to a load of 70% of the SMI,
if itis applied for a sufficient amount af time,
DESCRIPTION OF RESEARCH STUDY
A research program conducted by a consortium of 10 utilities and the Califomia Energy
Commission administered under the Electric Power Research Institute evaluated composite
instrument transformers at a commercial testing laboratory. Because of the research character of the
program, additional measurements, tests, and analyses were conducted to evaluate the composites
and the IEEE 693 standard used to qualify them. Some of these supplemental activities are
described below. ‘The configuration of an instrument transformer is illustrated in Figure 3. The
figure docs not show the 2.4 m (8 feet) support structure used in the test. It does show the location
‘of some of the accelerometers used during shake-able tests and the laser pointer location uscd
during the pull test. ‘The picture represcats a S00 kV unit that is fabricated from two composite
‘members, although some designs use more than two members.
Instrumentation
‘The instrumentation described below is related to the data discussed in the paper. Other
instruments not included are accelerometers a the center of gravity, accelerometers atthe top of the
support structure strain gages near the base of the support structure, and load bolis used to anchor
the support structure,
Laser Pointer
During the static pull tests a laser pointer was attached to the base of the column and positioned
horizontally to measure rotation at the base of the column. Rotation can be due to distortion of the
‘box lid, compression of the gasket between the box lid and box, and deformation of the box,
particularly near the anchor bolts. The laser pointer was projected on a vertical surface located a
distance from the column equal to the column height. As the cantilever load was applied in
increments, the deflection of the lascr spot was recorded, The rotation at the base contribuicd to the
deflection at the top that is not associated with insulator deformation. The measurement and
potential effects of the instrument transformer box lid rovation is not addressed in the IEEE 693
Standard,
‘Strain Gages on Bottom Flange
‘Two strain gages were mounted om the flange barrel, aligned parallel to the longitudinal axis, near
to the top edge of the flange. There were concers about using double integration of the
accelerations to estimate the deflection at the top. For this and other reasons discussed below, the
strain gages were added. They are shown in Figure 3. This is instrumentation not required by the
standard.
Accelerometers
“Three accelerometers are mounted at the top of the instrument-ransformer column measuring the
vertical and two horizontal accelerations. Two accelerometer are also located at the base of the
column that measure horizontal accelerations. Both seis of horizontal accelerations are used to
‘estimate the relative deflection of the column, These measurements are required by the standard.‘Figure 3 Mlustration of instrument transformer and instrumentation locations
TESTS AND ANALYSES NOT REQUIRED BY IEEE 693 STANDARD.
Man-Shake Test
Generally, after a large amplitude shake-able test a man-shake test was performed. This was
usually done by using 2 ladder on the shake table to guin access to the composite member and then
aitempt to shake it at its first resonant frequency ie get as large an amplitude as possible and then
allow the motion to decay. ‘This data was used to- estimate the damping and the damped-natural
frequency. As it tumed out, it was also used to identify bond degradation.
‘Static-Pull Test Following Shake-Table Tests
Because of concern that the acceptance criteria used by the standard, that is dynamic
deflection less than the 1/2 SML deflection, may not adequately characterize composite
performance, a pull test was done alter the vibration test. Since this was not part of the
standard, it was not clear what deflection results would constitute acceptance.Additional Analyses
‘As part of the research program, all test data was requested from the test laboratory and mast data
was provided. Strain gage and deflection data collected from the pull tests were evaluated and
‘compared to each other. This data was used to calibrate strain measurements from the compasite
member flange to deflections at the top and applied moment, Damped sine waves were fitted to the
man-shake response data to estimate damped natural frequencies and damping. Horizontal
acceleration data trom the composite member was double integrated to get displacements. A
sinusoidal top acceleration was assumed and closed-form expressions for the velocity and
displacement were developed and compared to data derived from numerical integration. This
model was used to evaluate the effect of cross talk between acceleration axes resulting from
accelerometer rotation associated with the bending of composite members. The deflection ebtained
from analog double integration of the top acceleration data and strain data during shake-table tests
were compared. Response data from 0.25g and 015g tests are compared and response data from
015g and Ig data are compared
ACCEPTANCE CRITERIA
‘There were several concerns associated with the acceptance criteria, First, the standard allows
double integration of acceleration to estimate relative deflection of the composite member, and this
‘was the method of choice of the testing laboratory. Any offset in the acceleration measurement will
‘cause the estimated displacement to get unrealistically large. The static-pull test and the first mode
dynamic response primarily intreduce bending in the column, Associated with this bending is a
rotation at the top of the column where accelerometers are located. Thus, an accelerometer
installed with its axis of sensitivity aligned in a horizontal direction when the composite member is
vertical, will have this axis rotate as the column bends. This results in cross-auis sensitivity so-that
as the column bends the accelerometer will pick up vertical acceleration that will be interpreted as
horizontal input. This will give an inaccurate measure of the top deflection, particularly when the
aceeleration data is integrated twice.
“The cantilever load in the static-pull test applies a moment at the base of the composite member.
‘The base moment is the critical variable that is mast likely to cause damage to the composite
member. The deflection at the top is a good measure of the base moment for this load. In the
dynamic test the load on the composite member is distributed over its length rather than
‘cancentraled at the top. so that the dynamic deflection will give a slightly higher base moment for
the same deflection at the top. Thus, this is a non-conservative estimate of the base moment. The
‘dynamic response will be influenced by the stiffness of the base connection and by the damping of
the system. A loss of stiffness due to degradation of the bond or purtial failure of mbe fibers will
tend to increase the response, so that the acceptance criteria may not be satisfied. However.
‘damping also inereases with damage and this will tend to reduce the response. It is not clear which
effect will have the largest influence. In shake-table tests higher vibration modes can be excited
and the deflection atthe top will not be an accurate measure of the moment at the base
Finally, rotation at the base of the composite member will contribute to the: top deflection. This
‘component of the top deflection is not associated with member distortion. This has the effect of
relaxing the residual deflection criteria.RESULTS FROM TESTS AND ANALYSIS AND THEIR INTERPRETATION
In the following discussion of test and analysis results, reference will be made to degradation of the
bond between the tube and flange. The degradation has been inferred by the interpretation of the
data, however, there has not been direct observation by the sectioning of the connection or by ultra
sound measurements.
Preliminary Statie-Pull Test
During the static-pull test of smaller diametcr, composite insulator instrument transformers, a loud,
energetic popping sound was observed on some units. In one case, as the load approached the 1/2
SML the output of the strain gage mounted on the flange that was in tension suddenly dropped
when the sound was heard. An investigation of the strain gage and the installation of a new strain
gage indicated that there was a drop in the strain on the flange at that location. An additional strain
‘gage was installed closer to the base of the flange and strain was observed. The popping sound and
loss of strain was interpreted to be duc to bond degradation. The extent of bond degradation is not
known. The presence of strain in the lower strain gage that was added may indicate the integrity of
the bond to that level, but this strain could also be observed without bonding.
‘The use of the laser pointer to measure rotation at the base of the composite member indicated that
10% to 15% of the peak deflection at the top was due to the rotation at the base. One unit that was
tested used a relatively thick cork gasket to scal the instrument-transformer box lid to the box, and
this increased the rotation at the base of the composite member. The increase in peak deflection has
the effect of relaxing the residual deflection criteria of the standard.
(On a large diameter composite instrument transformer, the strain gage output was scaled to
represent the deflection at the top by using one of the data points obtained during the static-pull
test. The scaled strain and the deflection plotied in Figure 4 show that these parameters track well
upto the 1/2 SML_
Pre-Test Disp and Strain, Pull Test. Large Diamete
a ee ee ee ee
Force ~ pounds
Figured Scaled strain and top deflection of initial static-pull testVibration Tests:
During the vibration tests of some composite instrument transformers, a loud popping sound was
‘heard. In one case, the euiput of accelerometers mounted on the composite member exhibited a
large spike, about twice the acceleration value of the largest amplitude ebserved during vibration
tests. This unit also expericnced popping during the initial static-pull test. An inspection of the
response before and after the pop did not show any marked difference in the performance.
During a test at the 1g level there was a sudden failure of a tube-flange joint, as shown in Figure 2.
Oil contained in the composite member ran from the joint. All indications of a test at 05g indicated
‘that the unit was performing well, This nonlinear characteristic of the failure means that testing 10
(05g and extending the results to 1g is inappeopriate for this type of equipment.
Figure 5 shows the superposition of the response of a large diameter, composite instrument
transformers during (025g and 0.5g tests. The 025g test was scaled by a factor of two in this plot.
‘This unit had not been subjected to an initial static-pull test. Most of the mismatch of the responses
js duc to a slight time: shift in the two mconts.
Large Diameter 1, Front-Back, Strain 5g & 2 x
600
i!
Hall
Strain
I Hi ah fi
o 5 10 15 20 25 30
Time ~ Second:
Figure 5 Scaled 0.253 and 0.5g response tracked well on unit not subjected to a stalic-pull test
It is interesting to compare the responses of two units subjected to (05g tests. Figure 6. At times the
‘ime histories track well and at other times the unit that was subjected to the pull test in the front-
‘back direction has a much smaller response. This is attributed to the mobilization of a degraded
‘bond in the unit that was subjected to the pull test. During the time-history response the bend
separation is mobilized and there is an increase in damping but when the response
they generally track better again. This is attributed to the shrink-fit compressive stress locking the
‘bond separation at low amplitudes.Large Diameter 1 & 2, Front-Back, Strain,
600,
i
A a
A
MI
Time - Secend:
Figure 6 Two identical units subjected to 0.5¢ input but unit that had a static-pull test at times
exhibits lower response that is attributed to increased damping
Strain
Finally, the response of a small diameter, composite instrument transformers to 05g and 1g are
‘with the same scales, Figure 7. Note that the responses are about the same
amplitude even thought the input of one is twice the Value of the other. It is felt that the bond of the
unit was severely damaged at the end of the 0.5g test or at the beginning of the Ig test. The results
‘of the man-shake test and final static-pull test are discussed below.
‘Man-shate tests
‘The man-shake test performed alter the shake-tble test on one composite instrument transformer
showed that at the largest amplitudes, which are about 10% of those observed during the 0.5g tests,
the fraction of critical damping was 0.02. However, as the amplitude decayed, the fraction of
critical damping suddenly changed to 0.0023. Figure 8 shows the plots in which a damped sine
‘wave was fitted to the response data. Plot A is the large amplitude part of the response und Plot &
continues where Plot A stopped. While it ix recognized that damping generally decreases as
amplitude decreases, the damping changed by a factor of 8 and the shift occurred during one cycle.
‘The interpretation of the change in damping is that at large amplitudes the portion of the degraded
‘bond was mobilized so that the fractured surfaces rubbed and increased damping. At lower
amplitudes the compressive stress associated with the shrink fit locked the fractured surface and
there was a sudden drop in damping.
‘The response ofa small diameter, composite instrument transformer that was subjected to Ig test
had a single damping value of 0.048 that was independent of amplitude. The authors believe that
the bond in this unit wus severely damaged and that accounted for the larger damping.Relative Deflection, Front-Back, Small Dia. 1, .5g & 19
Relative Deflection - inches:
0 5 10 1s 20 es 30
Time - Seconds:
Figure 7 The response of a unit to 0.5¢ and Ig are approximately the same
Finat Statie-Pull Test
‘The final static-pull test for the small diameter, composite instrument transformer that was
subjected to Ig input exhibited a loud grinding noise-as the load was applied. Just prior te reaching
the 1/2 SML, and there was a drop in load and an increase in deflection. The initiat and final pull
tests are compared in Figure 9 and show that the deflection of the final pull est is about 50%% higher
‘than the initial pull test. Using existing acceptance criteria this unit would have been qualified.
Figures 10 A and B show the comparison of the strain and deflection of the initial and final pull
{ests on a large diameter composite instrument transformer. While this unit appears to have
performed well when subjected to a 1g test, the increase in the strains and deflections suggests that
there was some bond degradation. The loss of some of the bond increased the strain om the
remaining bond area and the loss of stifiness due to bend degradation increased the relative
deflection.
RECOMMENDTIONS AND SUGGESTED CHANGES TO IEEE 693 STANDARD:
“The criteria for the applicability of the general procedure used in the IEEE 693 standard to test to
half of the performance level as a means to qualify to the projected performance level are not
satisfied for hollow-core composites. Potential of bond degradation, tube fiber damage and fiber
delamination cannot be directly obscrved or measured and these types of failures are nonlinear in
‘character. Thus, potential damage variables cannot be observed and evaluated to assure that the
values of the critical variables are less than the value observed during the 1/2 SML static-pull test.
‘Thus. it is recommended that hollow-core composites should be tested to the performance level.
“This reflects the character of the failure modes. While the results are based on the evaluation of &
small number of composite members, they clearly show the deficiencies in the standard more than
‘basic a problem with the inherent seismic capacity of this type of equipment.‘Acceleration = 9)
Man-Shake, Large Diameter 1, Front-Back, 10-17
Fn= 1.67, Z= .02,Amp=.27
10 " 12 13 14 15. 16 17
Time - Secend:
Figure 8 A The large response to man-shaking and a fitted damped sine wave
‘Acceleration - g
Man-Shake, Large Diameter 1, Front-Back, 17-30
Fn=1.7, Z =.0035, Amp= .09
Time - Secend:
Figure 8B The small response to man-shaking and a fitted damped sine wavePre & Post Pull Test Dispalcements - Small Diamet
20
Deflection Inches
a 100 200 m0 400 500 609
Fore - Pound:
Figure 9'The deflection in the final pull test is about 50% higher than the initial pull est
‘The acceptance criteria should consist of a static-pull test after the shake-tables tests are complete.
As an interim measure, the deflection at the 1/2 SML should be no more that 15% larger than that
‘observed in the initial pull test. Since differences in performance were observed in test data in the
different directions, a pull test should be conducted in both the front-back and the side-side
directions,
In each orientation, the pull test need only be performed once, and the residual deflection should be
less than 5% of the peak deflection. This suggestion is a departure from current practice, as it
would eliminate dynamic measurements during testing as part of the acceptance criteria and
rotation at base would not matter. Until rescarch substantiates the validity of the above tests, it is
recommended that a snap-hack vibration test to V8 SML be conducted to determine the extent of
‘bond degradation.
SUGGESTED RESEARCH
‘The acceptance criteria suggested above are based om 2 very limited research effort. The authors
feel that a static-pull test is needed after the shake-table testing is complete. It is anticipated that
there will he an increase in deflection at the 1/2 SMIL_as a result of the shake-table testing, however,
additional research is needed to establish the magnitude of the increase that can be accepied
‘without indicating the presence of unacceptable damage.
‘The test data suggests that the extent of bond degradation can be estimated by the level of motion
required to mobilize the degraded bond. Rescarch could establish the relation between bond
degradation and vibration amplitade needed to mobilize the joint. This could be used to determine
if the degradation was sufficient to jeopardize the integrity of the tube-flange seal. ‘The research
may also demonstrate that the increase in deflection during or the residual deflection measured in
the post-vibration static-pull testis adequate to assure seal integrity.Pre & Post Pull Test, Strain, Large Diamet:
é
i
z
2
é
o ee ee ee ee ee |
Foree - Pound:
Figure 10 A Relative strains increase in final statie-pull test
Pre & Post Pull Test Large Diameter 1, Displacer
12
4
i
&
a
° 100 200 300 400 500 600 700}
Foree - Pound:
Figure 10 B Relative deflections increase in final static pall testRescarch should be conducted to sce if bonding at the top part of the flange could be eliminated.
‘This could eliminate bond degradation and also increase system damping and thus improve the
dynamic response of the unit. As part of this effort, the stifliness af the flange could be investigated
to see if increased stiffness would reduce bond degradation.
“The test data suggests that the penctration length of the tube into the flange as a percentage of the
diameter should be investigated. Smaller diameter tubes appeared 1 have more problems with
‘bond degradation. It may be necessary that smaller diameter tubes have a larger penelration relative
to their diameter to mitigate bond degradation. Changes in pre- and post-test damping could be
used as a measure for establishing acceptance criteria of composite insulator bond integnty in high
voltage equipment. The test data showed changes in damping that could be the result of bond
degradation. Research should be conducted to determine the sensitivity of damping values relative
tw the level of hond degradation.
‘There is a need to assess the implications of popping sounds observed during tests on bond
degradation. The assumption that this only corresponded to bond degradation rather than bond
failure should be substantiated.
REFERENCES
1. IEE Standard 693, IEEE Recommended Practice for Scismic Design af Substations,
Institute of Electrical and Electronic Engineers, Inc, 345 Fast 47 Street, New York, NY.
10017,USA, 1997